WebPAC/Input/ISI.pm

package WebPAC::Input::ISI;

use warnings;
use strict;

use WebPAC::Input;
use base qw/WebPAC::Common/;

use Data::Dump qw/dump/;

=head1 NAME

WebPAC::Input::ISI - support for ISI Export Format

=head1 VERSION

Version 0.02

=cut

our $VERSION = '0.02';

=head1 SYNOPSIS

Open file in ISI export fromat

 my $input = new WebPAC::Input::ISI(
        path => '/path/to/ISI/records.txt',
 );

=head1 FUNCTIONS

=head2 new

Returns new low-level input API object

  my $input = new WebPAC::Input::ISI(
        path => '/path/to/ISI/records.txt'
        filter => sub {
                my ($l,$field_nr) = @_;
                # do something with $l which is line of input file
                return $l;
        },
  }

Options:

=over 4

=item path

path to ISI export file

=back

=cut

my $subfields = {
        'CR' => sub {
                my @v = split(/, /, shift);
                my $f;
                foreach ( qw/author year reference volume page/ ) {
                        if ( my $tmp = shift @v ) {
                                $f->{$_} = $tmp;
                        }
                }
                if ( $f->{author} =~ /^\*(.+)/ ) {
                        delete $f->{author};
                        $f->{institution} = $1;
                }
                return $f;
        },
};

sub new {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);

        my $arg = {@_};

        my $log = $self->_get_logger();

        open( my $fh, '<', $arg->{path} ) || $log->logconfess("can't open $arg->{path}: $!");

        my ( $format, $version );

        my $line = <$fh>;
        chomp($line);
        if ( $line =~ /^FN\s(.+)$/) {
                $format = $1;
        } else {
                $log->logdie("first line of $arg->{path} has to be FN, but is: $line");
        }

        $line = <$fh>;
        chomp($line);
        if ( $line =~ /^VR\s(.+)$/) {
                $version = $1;
        } else {
                $log->logdie("second line of $arg->{path} has to be VN, but is: $line");
        }

        $log->info("opening $format $version database '$arg->{path}'");

        my $tag;
        my $rec;

        $self->{size} = 0;

        while( $line = <$fh> ) {
                chomp($line);

                my $v;

                if ( $line =~ /^(\S\S)\s(.+)$/ ) {
                                $tag = $1;
                                $v = $2;
                } elsif ( $line =~ /^\s{3}(.+)$/ ) {
                                $v = $1;
                } elsif ( $line eq 'ER' ) {
                        # join tags
                        foreach ( qw/AB DE ID TI/ ) {
                                $rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_};
                        }
                        $rec->{'000'} = [ ++$self->{size} ];
                        push @{ $self->{_rec} }, $rec;
                        $rec = {};
                        $line = <$fh>;
                        chomp $line;
                        $log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' );
                } elsif ( $line eq 'EF' ) {
                        last;
                } else {
                        $log->logdie("can't parse +$. $arg->{path} : $line");
                }

                if ( defined $v ) {
                        $v = $subfields->{$tag}->($v) if defined $subfields->{$tag};

                        $log->debug("$tag: ", sub { dump( $v ) });
                        push @{ $rec->{$tag} }, $v;
                }

        }

        $log->debug("loaded ", $self->size, " records");

        $self ? return $self : return undef;
}

=head2 fetch_rec

Return record with ID C<$mfn> from database

  my $rec = $input->fetch_rec( $mfn, $filter_coderef );

=cut

sub fetch_rec {
        my $self = shift;

        my ( $mfn, $filter_coderef ) = @_;

        return $self->{_rec}->[$mfn-1];
}


=head2 size

Return number of records in database

  my $size = $input->size;

=cut

sub size {
        my $self = shift;
        return $self->{size};
}

=head1 SEE ALSO

L<http://isibasic.com/help/helpprn.html> is only sane source of document format which Google could find...
 
=head1 AUTHOR

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2007 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1; # End of WebPAC::Input::ISI
1	package WebPAC::Input::ISI;
2
3	use warnings;
4	use strict;
5
6	use WebPAC::Input;
7	use base qw/WebPAC::Common/;
8
9	use Data::Dump qw/dump/;
10
11	=head1 NAME
12
13	WebPAC::Input::ISI - support for ISI Export Format
14
15	=head1 VERSION
16
17	Version 0.02
18
19	=cut
20
21	our $VERSION = '0.02';
22
23	=head1 SYNOPSIS
24
25	Open file in ISI export fromat
26
27	my $input = new WebPAC::Input::ISI(
28	path => '/path/to/ISI/records.txt',
29	);
30
31	=head1 FUNCTIONS
32
33	=head2 new
34
35	Returns new low-level input API object
36
37	my $input = new WebPAC::Input::ISI(
38	path => '/path/to/ISI/records.txt'
39	filter => sub {
40	my ($l,$field_nr) = @_;
41	# do something with $l which is line of input file
42	return $l;
43	},
44	}
45
46	Options:
47
48	=over 4
49
50	=item path
51
52	path to ISI export file
53
54	=back
55
56	=cut
57
58	my $subfields = {
59	'CR' => sub {
60	my @v = split(/, /, shift);
61	my $f;
62	foreach ( qw/author year reference volume page/ ) {
63	if ( my $tmp = shift @v ) {
64	$f->{$_} = $tmp;
65	}
66	}
67	if ( $f->{author} =~ /^\*(.+)/ ) {
68	delete $f->{author};
69	$f->{institution} = $1;
70	}
71	return $f;
72	},
73	};
74
75	sub new {
76	my $class = shift;
77	my $self = {@_};
78	bless($self, $class);
79
80	my $arg = {@_};
81
82	my $log = $self->_get_logger();
83
84	open( my $fh, '<', $arg->{path} ) \|\| $log->logconfess("can't open $arg->{path}: $!");
85
86	my ( $format, $version );
87
88	my $line = <$fh>;
89	chomp($line);
90	if ( $line =~ /^FN\s(.+)$/) {
91	$format = $1;
92	} else {
93	$log->logdie("first line of $arg->{path} has to be FN, but is: $line");
94	}
95
96	$line = <$fh>;
97	chomp($line);
98	if ( $line =~ /^VR\s(.+)$/) {
99	$version = $1;
100	} else {
101	$log->logdie("second line of $arg->{path} has to be VN, but is: $line");
102	}
103
104	$log->info("opening $format $version database '$arg->{path}'");
105
106	my $tag;
107	my $rec;
108
109	$self->{size} = 0;
110
111	while( $line = <$fh> ) {
112	chomp($line);
113
114	my $v;
115
116	if ( $line =~ /^(\S\S)\s(.+)$/ ) {
117	$tag = $1;
118	$v = $2;
119	} elsif ( $line =~ /^\s{3}(.+)$/ ) {
120	$v = $1;
121	} elsif ( $line eq 'ER' ) {
122	# join tags
123	foreach ( qw/AB DE ID TI/ ) {
124	$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_};
125	}
126	$rec->{'000'} = [ ++$self->{size} ];
127	push @{ $self->{_rec} }, $rec;
128	$rec = {};
129	$line = <$fh>;
130	chomp $line;
131	$log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' );
132	} elsif ( $line eq 'EF' ) {
133	last;
134	} else {
135	$log->logdie("can't parse +$. $arg->{path} : $line");
136	}
137
138	if ( defined $v ) {
139	$v = $subfields->{$tag}->($v) if defined $subfields->{$tag};
140
141	$log->debug("$tag: ", sub { dump( $v ) });
142	push @{ $rec->{$tag} }, $v;
143	}
144
145	}
146
147	$log->debug("loaded ", $self->size, " records");
148
149	$self ? return $self : return undef;
150	}
151
152	=head2 fetch_rec
153
154	Return record with ID C<$mfn> from database
155
156	my $rec = $input->fetch_rec( $mfn, $filter_coderef );
157
158	=cut
159
160	sub fetch_rec {
161	my $self = shift;
162
163	my ( $mfn, $filter_coderef ) = @_;
164
165	return $self->{_rec}->[$mfn-1];
166	}
167
168
169	=head2 size
170
171	Return number of records in database
172
173	my $size = $input->size;
174
175	=cut
176
177	sub size {
178	my $self = shift;
179	return $self->{size};
180	}
181
182	=head1 SEE ALSO
183
184	L<http://isibasic.com/help/helpprn.html> is only sane source of document format which Google could find...
185
186	=head1 AUTHOR
187
188	Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
189
190	=head1 COPYRIGHT & LICENSE
191
192	Copyright 2007 Dobrica Pavlinusic, All Rights Reserved.
193
194	This program is free software; you can redistribute it and/or modify it
195	under the same terms as Perl itself.
196
197	=cut
198
199	1; # End of WebPAC::Input::ISI